How the theory can be tested in practice

Paper 8C · A research program for compound race pathology · Read on Zenodo

A theory is only worth having if it can be wrong in a way you can measure. This paper takes the clinical friction framework and turns it into five concrete trial designs that others can run, refine, disprove, or set aside. Each design fixes in advance what the framework predicts, and when it should count as having failed. It should be read as a hypothesis, not as a treatment manual, and it gives no instruction to the individual patient. It is an invitation to test it.

From picture to test

The picture is the same one the rest of this site runs on, the one I explain elsewhere: the brain and body run races, where several possible reactions run at the same time and the fastest one wins. Repeated choices carve tracks, so the same routes win more easily again. And pressure decides how fast the race gets settled. The pressure a clinician sees is not just the top layer's own. It is the sum of everything pushing up from below, through a column of layers: biology at the bottom, feelings in the middle, thoughts and behaviour at the top.

The sister paper (Paper 8B) took that picture and made a specific claim about a set of major chronic diseases: cancer, autoimmune conditions, post-viral fatigue syndromes, and severe depression that does not respond to treatment. The claim is that they share a compound mechanism, where several weak links across several levels add up, and that treatment aimed at a single target therefore comes up short again and again. If that is right, it points to a different kind of intervention: several targets at once, sorting patients by where in the column their pressure comes from, and an effort that does not let go too soon.

This paper asks whether that is just a good story or something you can test. The answer it gives is five trial designs, each one putting the framework to the test in a real clinical setting.

Five trials, one question

The five designs spread across three disease areas: psychiatry, the after-effects of a virus, and an autoimmune treatment. They were chosen because the infrastructure, the patients, and in several cases the data already exist, so each trial could be run, or at least piloted, within roughly one to three years.

• Ketamine plus structured follow-up care for severe depression. Ketamine briefly lowers the bottom of the column, so a window opens in which new, healthier routes can win. A structured course inside that window (CBT, cognitive behavioural therapy) is then meant to help the new track settle before the bottom rises again. The design is set against an existing trial (CBT-ENDURE) as the comparison. An open pilot study suggested that ketamine alone held for a few weeks, while ketamine plus therapy held for around three months.
• Psilocybin plus matched integration for severe depression. Same basic idea, but here proposed as a follow-on trial after any approval, where the follow-up care itself is tailored to the individual patient's profile in the column.
• Long COVID as a multi-track trial. Instead of testing one thing at a time, several interventions are tested in the same design, so you can see whether they work together. It is anchored in a signal that has already been observed: in a large trial (RECOVER-AUTONOMIC) a heart-medication intervention only worked when it was combined with multi-disciplinary care, and the interaction was statistically clear (p = 0.004). Patients are sorted by where their pressure comes from, so different subgroups don't get blended into one washed-out average.
• Profiling before the first treatment for depression. Here the patient's column is measured along several axes before the first medication is chosen, rather than trying one thing after another. It connects to a proposal for a standardised panel of biomarkers, which I come back to.
• CAR-T plus stabilisation afterwards for autoimmune disease. CAR-T resets part of the immune system forcefully and has produced drug-free remission in patients with severe lupus. The proposal is to support the bottom afterwards, so the new, more tolerant pattern gets the quiet it needs to settle. The Erlangen group around Schett is named as the obvious collaborator, because they already work in this area.

The shared question underneath all of them is the same: does it work better to hit several links in the chain at once, and to time the effort so the new track can settle, than to do one thing at a time? That is exactly what the compound framework predicts, and it is what the five designs are built to test.

What a trial design has to do

A loose proposal to "try several things" is not a test. What makes these five into real tests is that each of them fixes in advance what the framework predicts, how many patients it takes to see a difference if one exists, and when the result should count as the framework having got it wrong. That last point is the most important. A framework that cannot fail is not science.

The paper ties the five designs to a list of pre-specified conditions (called R1 through R10 in the sister paper), which each trial puts to the test. If a trial shows that the tailored, multi-track intervention does not beat the usual single-target effort, then it is not a defeat you can talk your way out of afterwards. That is how the framework was agreed it could fail. The paper also describes what trials of this kind should report, so that results can be compared and don't get lost in differences of method.

Same diagnosis, different roads there

One idea runs underneath all of it. A diagnosis like "treatment-resistant depression" or "long COVID" describes how it looks on the surface, not what created it. Two people can reach the same diagnosis by completely different roads: one through inflammation in the body, another through a stress system that has run off the rails, a third through the gut, a fourth through a deeply carved track of heavy thoughts.

That explains why sorting patients by a single measurement often fails. A trial that split depression patients by a single inflammation marker in the blood (PREDDICT) found no effect, and that is exactly what you should expect: the same marker can be high for many reasons, so it cannot on its own point to which road this particular patient took. The framework's answer is that you have to read several axes at once and see which pattern dominates. That is the difference between asking "which box does the patient go in" and "which of several possible roads created the condition in this particular patient".

A panel that has to prove itself first

One of the five designs connects to a concrete proposal: a panel of six axes that measures where a patient's pressure comes from in the column, before the first treatment is chosen. The six axes are inflammation, the stress system (cortisol), the gut's bacterial flora, thought patterns, a deficit in drive to act, and sleep. The thought is that two people with the same diagnosis can have completely different profiles, and that this explains why the same medication works for one and not the other.

But the paper is explicit about the order here. The panel is not a finished tool. Before it can be used to choose treatment at all, each axis has to show that it measures reliably: that it gives the same answer when you measure again, that it does not swing at random, and that it holds across laboratories. The cut-off values the paper mentions are starting points borrowed from neighbouring research, not settled numbers. That documentation is set up as a precondition, not as a result the paper claims to have. It is an honest marker that the idea comes before the proof.

What the framework adds, and what it can't do

A thread running through the rest of the clinical family applies here too: you can only add, not subtract. A track that has been carved cannot be carved away. So all five designs are about adding something: a lower bottom so the race isn't settled too fast, a competing route that can win instead over time, or the right timing so the new track gets to settle. It is the same logic, translated from a picture into something you can measure on real patients.

It is also worth saying what the paper deliberately does not do. It stays clear of treatment recommendations and of telling any individual patient what to do, and it does not require that the author is a clinician. It delivers five hypotheses in a form that researchers, trial units, and precision-medicine programmes can take and try out, sharpen, disprove, or ignore.

What it means (and doesn't mean)

This is an invitation, not a conclusion. The value lies not in the paper having proven anything. It lies in making an otherwise loose theory concrete enough that others can take it down off the shelf and try it against reality, with targets and criteria agreed in advance. The paper also collects no new data: it anchors the five designs in trials and measurements that already exist.

And it is explicitly neither a new diagnostic manual nor medical advice, and it does not replace a professional assessment of the individual person. If something here rings true about you or someone you care about, that is a conversation to have with a professional, not a conclusion to draw from a website.

What I don't know

There is no clinical co-author on the paper yet, and that is a real limitation: the trial designs were drawn by a theorist without clinical training in the disease areas they span, not tried or run in a clinic. No new data was collected either. The paper gathers existing findings and anchors the designs in trials and data that already exist, but none of the five trials has been run yet.

The biological claims about what sits at the bottom of the column are phrased as predictions, not as something that has been proven. The six-axis panel has to show that it measures reliably before it can be used, and that documentation is missing. The maths behind the framework is deliberately simplified: it shows the shape of the mechanism, not numbers you can compute further with. The next step is collaboration with people who can run the trials properly and decide whether the framework holds.

Read the paper

The full article is freely available on Zenodo (concept DOI 10.5281/zenodo.20059871):

Read on Zenodo → · Technical version · Dansk version

Related on this site:

• Paper 1 (Friction Theory) — the framework behind races and tracks that this research program builds on.
• The Learning page — the same mechanics from the other end: how tracks are built and settle.
• The Memory page — why dumping information does not teach.

The family of clinical papers (same framework, different angles):

• Paper 8 — One mechanism behind many diagnoses: the base framework, on pressure, tracks and experience, and why you treat at the bottom, not at the top.
• Paper 8B — Compound race pathology: the claim that several weak links in the same chain add up to disease over time.
• Paper 8D — Treat at the base: the preventive side, on cofactors below the diagnostic threshold.